Confessions of a New York Rupture Researcher: An Insider’s Guide and Critique

Abstract

The study of alliance rupture has become quite prevalent since 1990 and especially in the past 15 years where we have seen a noticeable surge in empirical publications on the subject. This honorary paper attempts to provide a critical review of this literature from the perspective of someone who has contributed to it in his collaborative work on a research program designed to investigate ruptures and to develop intervention and training models to resolve them. This paper is organized into three topics or sections: (1) alliance rupture, (2) rupture resolution, and (3) alliance training; and it addresses definitions, findings, questions, and lessons with regard to each topic. It suggests some clinical conceptualizations (concerning agency and communion as well as mutual recognition), training implications (regarding emotion regulation and deliberate practice), and methodological considerations (promoting pluralism and contextualism), along with future directions.

Many years ago, when I was just an adolescent, I took a class on autobiography and was introduced to Confessions of an English Opium-Eater by Thomas de Quincey. It is a nineteenth century autobiographical account of the author’s addiction and lived experience on opium. It is considered by many as a forerunner of the Gonzo literary movement that was popularized by Hunter S. Thompson in the 1970s. Symbolized by a raised fist and represented by the slogan “Let’s be honest,” Gonzo is a style of writing where subjectivity is privileged, where the author is part of the story, where the objective is “to tell it like it is” – typically with a critical and playful turn. Some believe is derived from the French-Canadian word gonzeaux, which means “shining light.” As my title suggests, I decided to base my address on the Gonzo sensibility and accordingly provide a critical look at a body of research which I have played a part of and lived through for almost 30 years. In other words, an insider’s guide and critique on rupture research. Over the past 30 years, we’ve seen a significant surge in published empirical studies on ruptures, especially in the last 10 years.¹So my thinking is: It is timely to turn a critical eye on what we’ve done, what we know and don’t know, and what we should do. This paper is organized into three parts –alliance ruptures, rupture resolution, and alliance training –and will address definitions, findings, questions and lessons that I’ve learned over the years.

What about alliance ruptures?

“Stuck in the moment you can’t get out of” – Paul Hewson (Bono from U2)

In many respects, the term rupture has been a challenging one: memorable and misleading. Although my colleagues and I have vacillated in our attachment to it, it is also something we became known for (e.g., Eubanks-Carter, Muran & Safran, 2010; Safran & Muran, 1996, 2000, 2006; Samstag, Muran & Safran, 2004): Recognizing this has always seemed to quiet any concerns. One definitional challenge is that there are so many terms and concepts associated with it. Here is a list of some, many of which have been used interchangeably: breaches (Safran, 1993), challenges (Agnew et al., 1994), disagreements, deteriorations and dysfluencies (Safran & Muran, 2000, 2006), disturbances, disruptions and misalliances (Langs, 1976), impasses (Hill et al., 1996), misattunements (Boston Change Process Group, 2010), [mis] coordinations (Westerman et al, 1995), misunderstandings (Rhodes et al., 1994), negative process (Binder & Strupp, 1997), strains (Bordin, 1994), threats (Bennett et al., 2006), and weakenings (Lansford, 1986). They have served to elaborate on the meaning of rupture, but also to add to some confusion. There are also a number of psychoanalytic concepts that overlap and have contributed to the confusion: for example, derailments in dialogue (Spitz, 1964), empathic failures (Kohut, 1984), enactments (Jacobs, 1986), resistance (Greenson, 1967), projective identification (Klein, 1975), transference-countertransference (Kernberg, 1965), vicious cycles (Horney, 1950). We have particularly used enactment and vicious cycle in efforts to define rupture and to emphasize the therapist unwitting participation (Safran & Muran, 2000, 2006).

We have been consistent, however, in defining rupture according to Edward Bordin’s (1979) transtheoretical reformulation of the alliance. He defined the alliance as purposeful collaboration (that is, agreement between patient and therapist on the tasks and goals of treatment) and affective bond (or emotional attunement between patient and therapist): see also Hatcher (2010) and Horvath (2017). Accordingly, ruptures can be understood at an explicit level as disagreements (in tasks or goals) and deterioration (in the bond). We also elaborated on Bordin’s reformulation by suggesting that at an implicit level these dimensions reflect an ongoing and underlying intersubjective negotiation between patient and therapist respective needs or desires (Muran, 2001, 2007; Safran & Muran, 2000, 2006). This elaboration brings therapist subjectivity into relief and suggests a more fluid dynamic between patient and therapist subjectivities, as well as an inevitability to ruptures as individual needs invariably clash. It also allows us to consider how we negotiate the oft-cited needs for agency and communion (Bakan, 1966; Benjamin, 1974; Beck et al., 1983; Blatt, 1990: I will develop this perspective further in this paper.

Over the years, there have been a number of attempts to operationally define ruptures using patient-report measures, based on or related to Bordin’s reformulation. They, of course, begin with the many studies and meta-analyses, suggesting low alliance indicates rupture (Horvath et al., 2011; our own contributions to this literature include Muran et al., 1995; Muran et al., 2009; Zilcha-Mano et al., 2016). There are also several studies using patient postsession reports that define ruptures in terms of sudden drops –with different formulas to define these drops. For example, a number have applied criterion-based methods (designating limits based on between-case analyses), defining ruptures by drops of one point on an alliance scale (Larsson et al. 2016; Stevens et al., 2007), of one standard deviation (Haugen et al., 2017; Strauss et al., 2006), or of .33 standard error of the difference (McLaughlin et al., 2014) in repeated measurement of overall alliance. In contrast, there have been a number of case-sensitive approaches based on within-case analyses: These include simply recording any drop in alliance rating from the previous session in a given case (Botella et al., 2008). In another example, Stiles et al. (2004) and subsequently Westra et al. (2011) used four parameters (intercept, slope, curve and variability) computed from regressions equations to define ruptures. We have recently applied a control chart method (Eubanks-Carter, Gorman & Muran, 2012; Lipner et al., 2016) that defines ruptures by deviations in the alliance relative to the individual patient’s own mean alliance rating, rather than a pre-established criterion: In this method, control limits comparable to 1.5 to 2 standard deviations have been use to define ruptures. Based on an exploratory cluster analysis of early alliance development (first four sessions), Zilcha-Mano and Errazuriz (2017) interpreted a rupture pattern –a downward trajectory in alliance ratings. There have also been two studies that have looked at a single index based on patient postsession report, directly asking, “To what extent did you feel any tension or problem, any misunderstanding, conflict or disagreement, in your relationship with your therapist during the session?” (Muran et al., 2009; Sommerfeld et al., 2008).

Table 1 presents the various efforts to operationally define ruptures using patient postsession report measures. In sum, eight studies indicated prevalence rates of ruptures in cases that range from 25% to 68%. Five of these studies also assessed predictive validity, four of which indicated statistically significant relationships to treatment outcome with mostly medium effects (Botella et al., 2008; Larsson et al., 2016; McLaughlin et al., 2014; Muran et al., 2009); three of the studies listed in Table 1 did not report on this relationship. It is important to note here that not all these studies assessed alliance after every session; some only assessed alliance early on; some included very brief treatments; and all the studies considered the alliance in a variety of psychotherapies: many cognitive-behavioral and some naturalistic treatment-as-usual.

Table 1.

Studies of patient post-session report of “rupture” and “rupture resolution”

Studies	Measure	Sessions used/ Treatment length	Sample Size	Rupture Prevalence	Rupture Prediction	Resolution Prevalence	Resolution Prediction
Stiles et al. (2004)	ARM	8/8 or 16/16	n=79	n/r	n/r	22%	medium*
Strauss et al. (2006)	CALPAS	8/52	n=25	68%	n/r	56%	med*-large**
Stevens et al. (2007)	WAI	30/30	n=44	n/r	n/r	50%	small-medium
Sommerfeld et al. (2008)	Single index	All/16–40	n=5	42%	n/r	n/r	n/r
Botella et al. (2008)	WAI	1st 8, every 4th/48	n=181	54%	small*	n/r	n/r
Muran et al. (2009)	Single index	1st 6/30	n=128	37%	medium**	n/r	medium*
Westra et al. (2011)	CALPAS	4/8	n=38	42%	small	16%	small
McLaughlin et al. (2014)	CALPAS	5/10	n=82	46%	medium**	28%	small
Larsson et al. (2016)	WAI	All/3–22	n=605	25%	medium*	15%	medium*
Lipner et al. (2016)	WAI	30/30	n=129	n/r	n/r	81%	medium*
Haugen et al. (2017)	CASF	All/M(SD)=26(28)	n=32	n/r	n/r	38%	small
Zilcha-Mano & Errazuriz (2017)	WAI	1st 4/4–31	n=166	51%	n/r	24%	n/r

^*p<.05

^**p< .01

^***p<.001

n/r = not reported

Note. CALPAS=California Psychotherapy Alliance Scale, WAI=Working Alliance Inventory (Horvath & Greenberg, 1989), ARM=Agnew Relationship Measure, CASF=Combined Alliance Short Form (Hatcher & Barends, 1996).

We have also found it useful to define ruptures by specific patient communications or behaviors –even though, we have always considered a rupture as something co-constructed by patient and therapist, as an interaction between their respective personality configurations and immediate needs (Safran & Muran, 2000). In this regard, we adopted the distinction between withdrawal and confrontation markers (Harper, 1988a, 1988b). See Figure 1 for illustrations. Accordingly, withdrawal ruptures involve movements away from self or other, efforts towards appeasement or isolation: They can also be understood as pursuits of communion at the expense of agency. Confrontation ruptures involve movements against other, efforts towards aggression or control: They can also be understood as pursuits of agency at the expense of communion.

Figure 1. Withdrawal and confrontation ruptures.

Copyright 2002 by Dalia Spektor, PhD. Printed with permission.

In our original studies (Safran, Muran & Samstag, 1994; Safran & Muran, 1996), we operationally defined ruptures using observer-based measures of interpersonal behavior and emotional involvement. Specifically, we applied the Structural Analysis of Social Behavior (SASB) and the Experiencing Scale (EXP) to define ruptures as involving avoidance, appeasement and dependence, control and criticism, and having limited to no emotional involvement. Since then, two observer-based coding schemes have been developed to more directly measure ruptures according to the withdrawal and confrontation distinction. The Rupture Resolution Rating Scale (3RS: Eubanks-Carter, Muran & Safran, 2015) defines withdrawal and confrontation markers as specific behaviors. For example, withdrawal markers include: denial, minimal response, abstract communication, storying telling/topic shift, deference and appeasement, content/affect split, and self-criticism/hopelessness. Confrontation markers include: patient rejects, defends, & controls; patient complains about therapist, therapy, parameters, and progress. The Collaborative Interactions Scale (CIS: Colli & Lingiardi, 2009) defines withdrawal as indirect ruptures (IDM) or “when the patient indirectly expresses a form of emotional disengagement” from self, other or process (p. 723); and confrontation as direct ruptures (DRM) or “when the patient acts in a hostile or sarcastic manner, complains heatedly about lack of progress, questions or rejects the tasks or goals of therapy” (p. 723).

Based on studies using these observer-based measures (Colli & Lingiardi, 2009; Eubanks et al., 2017), ruptures are seen as quite prevalent, more so than what is yielded from patient-report. Withdrawal ruptures on the 3RS (n=46) and CIS (n=32) are found in every session (100%). Confrontation ruptures in contrast are found in 43% (CIS) to 91% (3RS) of sessions. Only confrontation ruptures as measured by the 3RS have been shown to be predictive of outcome (with a medium effect). With regard to the SASB (Henry et al., 1986, 1990), negative process (or what we originally defined as rupture) was present 26% in sessions from low outcome cases (significantly more than in high outcome cases – with a large effect size).

Though we have made strides with regard to the study of ruptures, there are many questions that remain. Here are some: Does low alliance really mean rupture? What do patient postsession reports actually measure? Does patient-report measure alliance or compliance? (This question was raised in an early study when we found patients with friendly-submissive interpersonal problems reported better early alliances [Muran et al., 1994].) Are all ruptures the same? (Do they all have implications for outcome? Are some more important than others? Are some more iatrogenic, risk factors for adverse effects or deterioration in psychotherapy? ² And what about location? Are there differences between early versus late ruptures?). What about therapist-report and match between patient and therapist report? (We once conducted a study that demonstrated patient-therapist consistency in alliance ratings at the item level was more predictive than individual ratings [Rozmarin et al, 2007]; and more recently we have explored patient-therapist match with actor-partner interdependence analysis and found a predictive relation to outcome [Zilcha-Mano et al., 2016].) And what about therapist contributions and match between patient and therapist personalities, as well as cultural, racial or gender identities? (Too little has been done in this regard, but there is some increasing attention to cultural ruptures [Hook et al., 2017] linked to the notion of micro-aggressions [Sue et al. 2007].)

What about rupture resolution?

“A struggle for mutual recognition” – G.W. Friedrich Hegel

One of the definitional challenges regarding rupture resolution mirrors the debate regarding the alliance and outcome in general: Is it a precondition for change or a change process in and of itself? I think it fair to suggest that they are both true, especially if one believes in multiple change processes and if one recognizes the inextricable tie between technique and the therapeutic relationship that many have acknowledged (e.g., Strupp, 1989) and that is fundamental to Bordin’s reformulation. Another challenging question is: What are the critical interventions in the resolution of ruptures?

My colleague Jeremy Safran and I proposed a typology of rupture resolution interventions over 20 years ago (Safran & Muran, 1998, 2000). These were developed from our own extension observations and readings of psychotherapy process. The typology has gone through a few iterations over the years (e.g., Muran et al., 2010), and in its current conception it is comprised of: immediate repair strategies, which include (a) providing the rationale for tasks or goals, (b) clarifying misunderstandings, (c) changing tasks or goals, and (d) providing validation for in-session anxiety; and expressive resolution strategies, which include attempts to use a rupture marker as an opportunity to explore core relational themes. With regard to exploring ruptures, we have particularly promoted the technical principal of therapeutic metacommunication from Donald Kiesler (1996), which is essentially communication about the communication process as it unfolds (Safran & Muran, 2000). It can be understood as “mindfulness in interaction” and as a way of bringing intersubjective negotiation into relief, including the basic needs for agency and communion. It requires therapists to reflect on their own internal state as they explore the interpersonal field in the here-and-now.

To elaborate, rupture resolution can be understood as a struggle for mutual recognition that includes the expression and clarification of both patient and therapist subjectivities. This understanding of the psychotherapy process and especially rupture resolution is derived from Jessica Benjamin’s (1990) integration of German philosopher Hegel’s existential notion that we as humans are always caught or thrown in a struggle towards mutual recognition. Negation is normative in the human condition. (In this regard, some may think of Edward Tronick’s [2007] finding of the ubiquity of misattunement in his research on mother-infant interactions.) Accordingly, Benjamin suggested there are ongoing power plays between patient and therapist, accommodations and refusals to accommodate, which can be understood as ruptures and used to convey to the patient that the world is negotiable and comprised of others with separate subjectivities (see also Muran, 2007). Thus metacommunication can be understood as promoting the potential for mutual recognition –or to use a few like-minded notions, a meeting of minds (Aron, 1996), a moment of meeting (Boston Change Process Group, 2010), and a genuine conversation (Stern, 1998): There are others in the literature (e.g., Buber, 1958). Metacommunication can also be understood as promoting the potential for emotion regulation: Put another way, the simple act of putting feelings into words (especially before another) has been shown by emotion researches to have a regulatory effect on emotion (Gross, 2015).

To illustrate metacommunication and its potentials, consider this vignette based on an actual interaction: A young female patient begins a session all excited by something she saw while watching a recent Super Bowl half-time show: The Sri Lankan artist M.I.A. had flipped the bird in full view of the millions watching. The therapist says he was unaware, which leads to the following confrontation by the patient, “Yea, I should have expected this from you… You just don’t care about the things that matter… to me.” The therapist’s initial attempt to understand was cut off with a very definite “no!” After some anxious silence, the therapist acknowledges, “I’m really hesitant to say anything right now” –another attempt at metacommunication. Noticing a shift in the patient’s demeanor, a withdrawal, the therapist explores her immediate experience, and she expresses anxiety about the therapist’s possible reaction to her anger. The therapist acknowledges disappointing the patient, but also encourages her to talk more about her anxiety, which leads to her expressing embarrassment and fear that he will reject her because of her “strong” expression. The therapist eventually turns the exploration back to her disappointment in him. In response, the patient articulates her concern that at times he doesn’t recognize some things important to her, that he doesn’t take her seriously enough. She goes on to marvel at the “audacity” of M.I.A. and to express her wish and fear of being “an empowered woman.” The therapist recognizes her conflict plus how it may have emerged in the interaction they just had; the patient appears to agree –to complete a moment of meeting. In this interaction, the therapist comes to have a more enriched understanding of the patient; likewise, the patient comes to see the therapist in a more differentiated way. (See http://www.therapeutic-alliance.org/mixed-3.html for video demonstration.)

In a series of single-case studies, we operationally defined and confirmed rupture resolution as an expressive process in a task analysis that mixed quantitative and qualitative methods (Safran & Muran, 1996). Based on 16 matched sessions from eight cases of an integrative cognitive therapy, we found support for a stage process model in which resolution progressed from recognition and exploration to expression of the need for agency or communion. On the SASB, we showed a move to expression and affirmation. On the EXP, we showed an increase in emotional involvement. By these measures, we defined metacommunication –the exploration pf both patient and therapist construal of the rupture and the clarification of underlying needs –as integral to rupture resolution. This research has received a good deal of attention: For example, Safran & Muran (1996) and Safran & Muran (2000) have over 2000 citations according to Google Scholar. Though rich in yielding clinically-useful data (that is, “what to do when”), it is only based on a small sample size, and we need to replicate or conduct a large-scale verification study (a longstanding regret and ambition). There have been other task analyses of rupture resolution conducted by other researchers: Some have provided support for recognition and exploration (Agnew et al., 1994; Bennett et al., 2006; Swank & Wittenborn, 2013), others suggested the value of making extra-session links (Agnew et al., 1994; Bennett et al., 2006; Cash et al., 2014) or changing in-session tasks (Aspland et al., 2008). It is important to recognize that these studies included even smaller samples, applied different methods, and involved a variety of treatment models –from dynamic, cognitive-behavioral to humanistic-experiential. In her review of qualitative studies of negative process, Hill (2010) also promoted the value of recognition and exploration.

In addition to this support at an intensive micro-analytic level, there is a growing body of evidence regarding rupture resolution, based on patient post-session report, specifically defining V-episodes, from sudden drops to a return of some norm and using different formulas or methods as mentioned previously. Table 1 also presents the various efforts to study rupture resolution in this regard. In sum, these studies indicate prevalence rates of such episodes in cases that range from 22% to 81%. Five of nine studies also indicated statistically significant predictive relationships to outcome with mostly medium effects (Muran et al., 2009; Larsson et al., 2016; Lipner et al., 2016; Stiles et al., 2004; Strauss et al., 2006).

With regard to observer-based methods, there is less evidence regarding rupture resolution and outcome. Our task analytic effort (Safran & Muran, 1996) was based on a method that indirectly related our stage process model to outcome –that was founded on an assessment structure linking observed in-session events to overall outcome (see Muran, 2002, for an explanation). Only Eubanks et al. (2017) applying the 3RS demonstrated a statistically significant moderate relationship of a single index of resolution to outcome (p < .05): Third-party raters rated, “To what degree were ruptures resolved over the course of the session.”

Of the many questions that remain regarding the study of rupture resolution, here are some obvious ones: What are we really demonstrating by between-session differences or V-episodes based on patient postsession report? Also, remember in this regard some of these studies did not assess after every session (McLaughlin et al., 2014; Strauss et al., 2006). And should we expect to see resolution within a single session or across multiple sessions. What are the critical processes? For example, can we conclude from all the task analyses taken together that just rupture recognition is essential? Or are there multiple resolution processes or pathways, depending on treatment orientation or on the dyad? These all speak to how much more exploration, verification and replication is needed.

What about alliance training?

“Deliberate practice” – Anders Ericsson

A natural consequence of the research on alliance and rupture is the development and evaluation of alliance training models. One definitional challenge is: What is the difference between alliance training and good training in general? An obvious answer is that the former is more focused on the nature and quality of the therapeutic relationship. However, I’m not confident all would agree with this, and sometimes I’m not even sure where I stand on this. Another challenge is what are the critical skills and effective strategies? I suppose we can agree rupture recognition is important, but how do we develop recognition skills –and what else is critical and effective?

I begin this section by reviewing the many studies that tested the impact of alliance training models on interpersonal process and treatment outcome. Table 2 provides a list: Many of these studies have explicitly incorporated principles from our work on rupture resolution, such as the studies from Penn State University (Castonguay et al, 2004; Constantino et al, 2008; Newman et al., 2011), University of Pennsylvania (Crits-Christoph et al., 2006), University Massachusetts at Amherst (Smith-Hansen et al., 2011), as well as the OQ System developed at Brigham Young University with its clinical support tool (e.g., Shimokawa, Lambert & Smart, 2010). As Table 3 suggests, many demonstrated some positive effect on interpersonal process or treatment outcome, but some had qualifications (Bien et al, 2006; Crits-Christoph et al., 2006; Smith-Hansen et al., 2011). A meta-analysis we conducted indicated small-to-medium between-group effects and medium-to-large within-group effects (Safran, Muran & Eubanks-Carter, 2011). On balance, the evidence is mixed.

Table 2.

Studies of alliance-focused training

Study(ies)	Process	Outcome	Summary
Vanderbilt U: Bien et al., 2006; Henry et al., 1993		+*	Training in time-limited dynamic therapy with focus on negative process. Mixed results regarding process and outcome (*significant medium effect only at follow up).
Penn State University: Castonguay et al, 2004; Constantino et al, 2008; Newman et al., 2011	+*	+	CBT training with additional focus on rupture-repair. Significant large effect on outcome (one study on depression); medium effects on patient-rated alliance and outcome (another study on depression); small to medium effects on patient-rated outcome (study on GAD).
University of Pennsylvania: Crits-Christoph et al., 2006	+*		Alliance-focused training (with rupture exploration): Medium effect on patient rated alliance (*nonsignificant); small effect on outcome (regarding depression).
University of Queensland: Bambling et al., 2006	+	+	Two alliance-focused supervisions (skill versus process focus): Significant medium effects on patient-rated alliance and outcome
OQ System: Shimokawa, Lambert & Smart, 2010		+	Feedback to therapists regarding alliance and with clinical support tools (with rupture-repair focus): Significant medium effect on preventing treatment failure
University of Massachusetts at Amherst: Smith-Hansen et al., 2011	+*		Video-assisted alliance workshop (with rupture-repair focus): Mixed results with medium relation of alliance strategies to therapist-rated alliance and patient attendance (*not patient- rated alliance or treatment outcome)
Mount Sinai Beth Israel: Muran et al., 2005; Muran et al., in press; Safran et al., 2005	+	+	Alliance-focused training with rupture-repair focus: Significantly better retention than CBT and STDP (medium effect); significant medium to large effects on interpersonal process (more positive, less negative).

Note. “+” indicates some positive findings.

Table 3.

Therapist narrative responses to post-training supervision survey on AFT impact

Therapist A: “Although I still felt somewhat stuck in sessions and my patient still seemed to respond poorly to my interventions, supervision helped me feel more free to try to express my thoughts and feelings to my patient.”

Therapist B: “We talked about my emotions a lot, so I was able to explore my emotional reactions… and learned some of my own limitations as a therapist.”

Therapist C: “I allowed myself to accept and tolerate the negative feelings I had felt towards my patient at times. This allowed me to gain a better understanding of what was happening in the therapy and to use that understanding in treatment.”

Therapist D: “Helped me not to hold things back, that my reactions can be used therapeutically.”

Therapist E: “Supervision allowed me to articulate and express a difficult internal reaction I had been having to my patient’s emotional distance. Through a two-chair exercise I had the opportunity to experience that my fears of expressing negative feelings towards my patient for fear of hurting him did not sound as harsh when I actually said them. This gave me the confidence to share these thoughts with my patient in session… This allowed me to more freely express myself in session… with more spontaneity and less deliberation.”

Therapist F: “Lots of role plays in which to explore techniques and areas which I had not experienced before. Also, pointing out ruptures that I was not aware of… I was under the impression it was not okay or therapeutic to include my own reactions and I learned how to do so in a manner which could be useful to the patient.”

Therapist G: “I was afraid of my patient’s anger… Through supervision, I was able to use this fear. I expressed a portion of my experience of fear using metacommunication. This freed me of the constraints fear had imposed on me and allowed the patient to move deeper into his experiences of us as a dyad.”

Note. These narratives are from therapists whose patients were in the top quartile for demonstrating the most change towards expressiveness (n=10/40: Three did not complete the survey).

We have conducted two studies funded by the National Institute of Mental Health (NIMH) that have contributed to this literature. Our first study (Muran et al., 2005; Safran et al., 2005) aimed to develop an intervention and training protocol (Safran & Muran, 2000) primarily informed by our stage-process model of rupture resolution (Safran & Muran, 1996). The efficacy of the intervention was evaluated compared to standard treatments on samples of patients diagnosed with a personality disorder (Muran et al., 2005) and determined at risk for treatment failure (Safran et al., 2005). The results suggested some promise for the intervention, especially with regard to treatment retention. This was a small-scale study that involved (what we thought was) a cross-over design and included patients who had also demonstrated difficulty in establishing an alliance with a previous therapist (see Muran et al., 2005; Safran et al., 2005). Our plan was to request further funding from NIMH for a larger-scale study based on this design: Regretfully, the program officer was not encouraging; the players or culture at NIMH had changed; so instead, following discussions with him, a second proposal emerged.

Our second NIMH-funded study (Muran et al., in press) aimed to further evaluate the training protocol developed in the previous study –specifically with regard to its impact on a cognitive-behavioral therapy (CBT). It consisted of two phases (see Figure 2). The first involved training novice therapists (doctoral candidates in clinical psychology) to an adherent standard in a 30-session protocol of a CBT for personality disorders. The second phase involved introducing our alliance-focused training (AFT) following a multiple baseline design (mixed within-subject and between-group): In this phase, our AFT was introduced to one cohort of therapists after eight sessions and another after 16 sessions. Our dependent variable in this study was interpersonal process (namely therapist and patient interpersonal behaviors) as measured by an abbreviated version of the SASB. Our aim by this mixed design was to try to control for variance attributable to patient, therapist and parient-therapist interactional sources that we thought undermined strictly between-group designs.

Figure 2.

Multiple baseline design of training conditions

In brief, our training model targeted the following therapist skills (see Muran, Safran & Eubanks-Carter, 2010): At the explicit level, we aimed to develop rupture recognition and resolution abilities (as described above); at the implicit level, we aimed to develop interpersonal sensitivity (greater orientation to process) and emotion regulation (greater facility in the negotiation of negative emotions). We employed the following training strategies: a didactic component comprised of manualization and videotape analysis and an experiential component comprised of mindfulness training and awareness exercises, including role-plays and two-chair work. With these components, our AFT places a great emphasis on clearly defined practice regimens that include pressured simulations and continuous feedback –what can be considered a form of deliberate practice (Ericsson et al., 1993).

In sum, we found the following interaction effects that demonstrated the impact of our AFT on interpersonal process. With regard to therapist behaviors, we found increases in affirmation and expressiveness and decreases in criticism and control, attributable to AFT. With regard to patient behaviors, we found increases in expressiveness and decreases in appeasement and dependence, attributable to AFT. Figure 3 illustrates interactions with regard to AFT’s facilitation of therapist and patient expressiveness, which could be interpreted as suggesting the possibility of mutual recognition. These findings mirror what we defined and confirmed with our stage process model from our original task analysis (Safran & Muran, 1996). These behaviors were found to be related to outcome in previous studies (e.g., Henry et al., 1986; 1990), but we also demonstrated the relation of many of these to growth curves of session impact, as well as to a composite measure of ultimate outcome (see Muran et al., in press).

Figure 3.

An illustration of interaction effects demonstrating the impact of AFT on interpersonal process (and mutual recognition)

To explore the impact of our AFT on therapist expressiveness further, we conducted two additional ancillary analyses. First, we assessed the effect of AFT on therapist emotional involvement based on the application of the EXP on a therapist interview conducted at treatment termination and that concerned the nature of the therapeutic relationship (see Safran et al., 2014 for preliminary analysis). We analyzed between-group differences (AFT Training at Session 8 versus 16) on EXP ratings of the interview (n = 30) and found greater emotional involvement with greater exposure to AFT, medium-to-large effect (t[28]=2.38, d=.90, p = .023). We also examined therapist descriptions of the impact of our AFT from narrative responses to a post-training supervision survey. Table 3 presents narratives from therapists whose patients showed the most change in expressiveness (top quartile). These narratives provide some insight to (more contextually rich data on) the training process and further underscore the importance of the therapist experience and expression. From these narratives, it appears freedom and support were critical themes for our trainees: By freedom, I mean from certain fears or anxieties; by support, I refer to validation of experience and encouragement to use it. Though based on a limited sample, I believe these narratives bring to life the importance of agency and communion in the training process, as well as in the treatment process. These data also remind me of relational theorist Neville Symington (1982) who suggested that we always struggle to find an act of freedom to break from the grip of an impasse or enactment: Metacommunication can be understood in this way –and our AFT as facilitative.

What questions remain for alliance-focused training? Here are some worth considering: What are the essential elements? (This suggests dismantling studies: For example, with regard to our AFT, could just mindfulness training for therapists be effective?) How do we evaluate competence on such principles as metacommunication –on therapist abilities to reflect-in-action (Schon,1984), as well as communicate and explore? Are there other ways to train or disseminate principles derived from alliance-focused research to clinicians? (For example, the OQ System developed by Lambert and colleagues [see Shimokawa, Lambert & Smart, 2010] includes principles derived from our work in their Clinical Support Tool, and yet it is not clear from their feedback research if and how this tool is actually used by clinicians. We are also interested in developing a video-enhanced version of this tool to improve the quality of the feedback.) And what can we learn from performance science (e.g., how do we translate such notions as deliberate practice to alliance-focused training?), cognitive science (what can we learn from research on attention, appraisal, decision-making under stress or uncertainty?), and emotion science (what can we provide therapists from the emotion science literature to negotiate their own experience of such negative states as: anxiety and panic, anger and frustration, sadness and despair, or shame and humiliation?) There are many disciplines we can learn from and many things I/we believe in and have yet to demonstrate.

Lessons learned

I would like to outline some of the lessons I have learned over the years – many have been suggested in what I have presented thus far.

Clinical.

To begin, here are some clinical lessons that apply to practice and training, which I have come to really appreciate:

• Be curious, court surprise and genuine conversation:

Here I am reminded of Jerome Bruner’s (1962) notion of “effective surprise” as well as ErnstSchachtel’s (1959) “allocentricity” and Shunryu Suzuki’s (1970) “beginner’s mind.” These suggest a sensibility open to difference or other. Most of all, in this regard, I have been influenced by Donnel Stern (1997), who described a psychotherapeutic process based on a dialogic epistemology (Gadamer, 1975). Accordingly, understanding (including self- and other-awareness) can only be more fully realized through “a genuine conversation” with another, where preconceptions (thoughts and beliefs) can be revealed in contrast to those of the other. The orientation is towards defining difference (as well as similarity) in a more nuanced way. I believe this is true of supervision too.

• Mind and mine the process:

During the 1992 U.S. presidential campaign, Bill Clinton’s staff supposedly kept their focus on the simple message, “It’s the economy stupid!” I remember Marvin Goldfried playfully borrowing from this at an SPR meeting with the observation, “It’s the relationship stupid!” It impressed me that this pioneer of the cognitive-behavioral tradition, as well as the integration movement, would place such emphasis on the therapeutic relationship. What I came to further appreciate was the importance of immediate process, the power of the moment, in the therapeutic relationship. It is not enough to see this relationship as ground for technique or as reflection of personal dynamics, but the potential for change is maximized by the recognition of relational experience as it unfolds. This is a process-experiential orientation (Greenberg et al., 1993) with a relational twist (Safran & Muran, 2000): Accordingly, agency can be courted as one observes an experience emerge in the therapeutic relationship, which in turn can transform the experience of the relationship: For example, as one recognizes an impulse emerge, one can choose to act or not on it; and even the recognition alone can make acting on the impulse different.

• Embrace emotion regulation (in therapy and training).

Research on emotion regulation by emotion scientists has flourished since the 1990s, demonstrating its adaptive significance, as well as the regulatory value of various cognitive reappraisal strategies (see Gross, 2015). There are a couple treatment models that have been developed that explicitly target it with the application of mindfulness training for patient dysregulation (Linehan, 1993; Menin & Fresco, 2014). There has not been much empirical study of emotion regulation for therapists in the context of therapy and training. Jon Kabat Zinn’s (1990) famous cognitive reappraisal “full catastrophe living,” takes on a different meaning when considered from the perspective of the therapist. In this regard, I encourage myself and my supervisees withthe reappraisal “pressure is a privilege” from the tennis great Billie Jean King (http://www.azquotes.com/quote/782738): We are doing important work and nothing is more important than how we negotiate ruptures.

• Practice deliberately and under pressure:

Anders Ericsson (Ericsson et al., 1993) introduced the notion of “deliberate practice” to describe the process towards expertise or mastery. It has been described as a well-defined practice regimen that is based on effective techniques, requires full cognitive commitment, stretches the individual beyond current abilities, involves expert- and self-monitoring, and produces effective mental representations. Others (e.g., Beilock, 2010) have contributed to this perspective with emphasis on the importance of “practicing under pressure.” It is useful to also consider Lev Vygotsky’s (1978) notion of “the zone of proximal development” in this regard: that is, the dimension between what one can do without help and what one cannot do (see Leiman & Stiles 2001). The literature on performance science supports the value of guided role-play exercises in the training of psychotherapists, as well as the use of videotape for feedback. In this regard, I have always appreciated the qualification from the American coaching icon Vince Lombardi, “Practice does not make perfect. Only perfect practice makes perfect” (http://www.quotery.com/authors/vince-lombardi/).

• Gauge safety and anxiety:

Though important to stretch someone beyond their comfort zone, it is also important to remember there is an optimal level of anxiety for communication and performance. Harry Stack Sullivan (1968) captured this notion with his “gradient of anxiety.” John Bowlby (1988) “secure base” can also be considered in this regard. For therapists and supervisors, it is critical to create a context where the patient or supervisee feels “safe enough” to explore and express their feelings, their impulses and fears. At an explicit level, this includes efforts to continually track their experience (respecting needs for privacy) and grant them agency in managing an exploration. At an implicit level, this includes a reliable and supportive stance by the therapist or supervisor. Although supportiveness and expressiveness have been characterized on a continuum (Wallerstein et al., 1956), I see these as separate dimensions that can be maximized at once: In other words, with sufficient support, there can be greater expressiveness –not a unique perspective (e.g., Luborsky, 1984).

• Appreciate individual growth curves (for patient or supervisee):

Everyone “[marches] to the beat of [their] own drum,” as Henry David Thoreau (1854) poetically put it. Everyone has their own developmental trajectory. Each patient and supervisee comes with their own unique history, their own individual configuration of dispositions and experiences, reactions and capacities. The therapist and supervisor must try to recognize the implications of this in a given moment and with regard to whatever challenge is being faced. Again appreciating difference is the orientation. We must try to help our patients and supervisees find their own unique solutions to their problems. This must be founded on a recognition of and respect for their current abilities. Implicit to this is the challenge of acceptance and hope that whatever one can do is “good enough” for the other (Winnicott, 1971).

• Monitor supervisory alliance ruptures as well.

Practice what you preach or “walk the talk” –a modernization of St. Jerome’s old adage. Teaching and learning can be understood as didactic, illustrative, experiential and vicarious. As supervisors, we should try to marry these modalities. At an explicit level, this includes playing your own video-demonstrations, participating in role-plays, and addressing ruptures as they occur in the supervision process. At an implicit level, this includes a consistent “walk” with regard to curiosity and openness, awareness and responsibility.

Research.

Here are some research lessons that I have learned and hopefully illustrated in this paper:

• Be complex:

I have always been wary about what gets “lost in translation” from concept to method. Concepts like alliance and rupture are complex in definition –and thus should not be simply reduced to patient-report as one example. So try to be sophisticated, comprehensive, pluralistic in method: Consider multiple reports –patient, therapist and observer. Remember the metaphor of the three blind men touching an elephant: They can each describe an aspect of an overall truth. And apply mixed methods –from quantitative to qualitative. Work at the global and local levels, back and forth from the aggregate to the single case, to realize a contextualist perspective and provide “thick descriptions” (that is, of behavior in context) as the anthropologist Clifford Geertz (2014) advocated.

• Be creative:

Think outside of the psychotherapy research box! Buck or build on convention! Consider alternative designs and analyses. Study and learn from other research disciplines. For example, with regard to time-series, we have experimented with methods (designs and analyses) from behavioral analysis (e.g., multiple baseline designs), quality control in industry (e.g., control charting), economic forecasting or climate change (e.g., alternative multi-level modeling strategies). Another good example seminal in psychotherapy research is the adoption of the task analytic method from cognitive psychology (Rice & Greenberg, 1984).

• Be specific and humble:

Remember to be contextually sensitive and critical. Be careful of absolute or over-reaching conclusions (which I probably violated by my interpretation of possible mutual recognition regarding the increases in patient and therapist expressiveness as a result of our AFT). Recognize, “The devil’s in the details” (or as the architect Mies Van der Roh [1959] put it more positively, “God is in the details”). When we identify V-episodes based patient-report, for example, we should probably be more measured about considering them to be rupture resolution events, especially given the general nature of the applied questionnaire, its administration post-session, and that in some studies assessment did not take place after every session.

• Be politic:

Remember research is a sociopolitical process: There is so much that goes behind the scenes in managing a research program –from relationships to administrators, supervisors, therapists, patients, research collaborators and assistants. There are many stakeholders with competing interests in the research endeavor. It is an interpersonal challenge as much as an intellectual one. Know your audiences: Try to hear their respective needs. There are so many stories that could be told in this regard, but here I will only highlight the challenge of convincing therapists participating in our research program at Mount Sinai Beth Israel that it had become important to study them as more than something synonymous with the treatment condition.

• Be collaborative & generative:

Recognize the importance of teamwork: “Stronger together” is not “an alternative fact” (to borrow from the current U.S. political discourse, as well as George Orwell). I am a devout believer that understanding and creativity are sown in dialogue and collaboration with others. Promote your students (e.g., give them first authorships; encourage them to publish alone): You grow from their growth; they are your future colleagues, and they carry the baton.

With this last lesson in mind, let me acknowledge the many who helped meet my needs for agency and communion: the many mentors, models, supervisors, colleagues and collaborators; the too many students to mention; the many patients I cannot mention; and of course my wife and son –plus my circle of friends and family. I also want to acknowledge the many other influential ideas: I began this paper with a reference to the gonzo sensibility and a raised fist as its symbol for the pursuit of truth. Let me close by invoking the symbol of the open hand and eye of knowledge (the Hamsa or Hand of Fatima) that my parents introduced to me as a child. For me, they both are true –and further underscore the significance of many, the value of pluralism, as well as the significance of history, the relevance of contextualism, which are fundamental to my approach to theory, research and practice (see also Muran, 2002). As an old supervisor once told me (not sure if this is his), there are many roads to Rome –and Paris is a beautiful city too!

Clinical & methodological significance statement.

The paper provides a critical review of research on a possible risk factor and potential change process in psychotherapy, including related training strategies, that has important implications for treatment outcome and professional development.